SOH Estimation of Li-Ion Batteries Based on Broadband Impedance Measurements and Equivalent Circuit Model Analysis

Lithium-ion (Li-ion) batteries have become the state-of-the-art technology in transportation and energy storage applications. In these applications, the high energy and power levels introduce significant challenges for the battery management system (BMS) that guarantees the safety of the battery system. In addition, the BMS monitors the state of the battery in terms of state-of-charge (SOC) and state-of-health (SOH), which are important parameters for the operation of the battery system.

A significant challenge for the operation of the BMS is introduced by the aging of the battery cells, which slowly degrades the performance of the whole battery system. As a result of aging, the capacity of the battery degrades which is monitored by the SOH. The capacity of the battery is difficult to obtain directly, so indirect methods are required to estimate the battery capacity in practical applications. Recently, battery internal impedance has been recognized as an effective quantity to access the capacity and power fade of the battery cell. Utilization of the impedance for the SOH estimation requires impedance measurement and modeling techniques to access the relevant aging information of the battery. For practical applications, these techniques should be implemented on- board the battery system, which is challenging due to the limited hardware of the battery system BMS and the operating restrictions given by the battery application.

This thesis presents methods for impedance-based SOH estimation of Li-ion batteries that are suitable for on-board implementation. The presented methods include broadband techniques to measure the battery impedance and modeling techniques to obtain parameters from the measured impedance revealing different aging mechanism and capacity fade of the battery under test. The presented impedance measurement, modeling and SOH estimation methods are validated by a number of experimental results. The results show that the presented methods have attractive on-board implementation properties that can be well utilized in practical battery applications.